===================================================================
RCS file: /cvs/mandoc/mdoc.c,v
retrieving revision 1.130
retrieving revision 1.179
diff -u -p -r1.130 -r1.179
--- mandoc/mdoc.c	2010/05/13 06:22:11	1.130
+++ mandoc/mdoc.c	2011/02/06 22:05:20	1.179
@@ -1,6 +1,7 @@
-/*	$Id: mdoc.c,v 1.130 2010/05/13 06:22:11 kristaps Exp $ */
+/*	$Id: mdoc.c,v 1.179 2011/02/06 22:05:20 kristaps Exp $ */
 /*
- * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
+ * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the above
@@ -21,72 +22,16 @@
 #include <sys/types.h>
 
 #include <assert.h>
-#include <ctype.h>
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <time.h>
 
+#include "mandoc.h"
 #include "libmdoc.h"
 #include "libmandoc.h"
 
-const	char *const __mdoc_merrnames[MERRMAX] = {		 
-	"trailing whitespace", /* ETAILWS */
-	"unexpected quoted parameter", /* EQUOTPARM */
-	"unterminated quoted parameter", /* EQUOTTERM */
-	"argument parameter suggested", /* EARGVAL */
-	"macro disallowed in prologue", /* EBODYPROL */
-	"macro disallowed in body", /* EPROLBODY */
-	"text disallowed in prologue", /* ETEXTPROL */
-	"blank line disallowed", /* ENOBLANK */
-	"text parameter too long", /* ETOOLONG */
-	"invalid escape sequence", /* EESCAPE */
-	"invalid character", /* EPRINT */
-	"document has no body", /* ENODAT */
-	"document has no prologue", /* ENOPROLOGUE */
-	"expected line arguments", /* ELINE */
-	"invalid AT&T argument", /* EATT */
-	"default name not yet set", /* ENAME */
-	"missing list type", /* ELISTTYPE */
-	"missing display type", /* EDISPTYPE */
-	"too many display types", /* EMULTIDISP */
-	"too many list types", /* EMULTILIST */
-	"NAME section must be first", /* ESECNAME */
-	"badly-formed NAME section", /* ENAMESECINC */
-	"argument repeated", /* EARGREP */
-	"expected boolean parameter", /* EBOOL */
-	"inconsistent column syntax", /* ECOLMIS */
-	"nested display invalid", /* ENESTDISP */
-	"width argument missing", /* EMISSWIDTH */
-	"invalid section for this manual section", /* EWRONGMSEC */
-	"section out of conventional order", /* ESECOOO */
-	"section repeated", /* ESECREP */
-	"invalid standard argument", /* EBADSTAND */
-	"multi-line arguments discouraged", /* ENOMULTILINE */
-	"multi-line arguments suggested", /* EMULTILINE */
-	"line arguments discouraged", /* ENOLINE */
-	"prologue macro out of conventional order", /* EPROLOOO */
-	"prologue macro repeated", /* EPROLREP */
-	"invalid manual section", /* EBADMSEC */
-	"invalid section", /* EBADSEC */
-	"invalid font mode", /* EFONT */
-	"invalid date syntax", /* EBADDATE */
-	"invalid number format", /* ENUMFMT */
-	"superfluous width argument", /* ENOWIDTH */
-	"system: utsname error", /* EUTSNAME */
-	"obsolete macro", /* EOBS */
-	"end-of-line scope violation", /* EIMPBRK */
-	"empty macro ignored", /* EIGNE */
-	"unclosed explicit scope", /* EOPEN */
-	"unterminated quoted phrase", /* EQUOTPHR */
-	"closure macro without prior context", /* ENOCTX */
-	"no description found for library", /* ELIB */
-	"bad child for parent context", /* EBADCHILD */
-	"list arguments preceding type", /* ENOTYPE */
-	"deprecated comment style", /* EBADCOMMENT */
-};
-
 const	char *const __mdoc_macronames[MDOC_MAX] = {		 
 	"Ap",		"Dd",		"Dt",		"Os",
 	"Sh",		"Ss",		"Pp",		"D1",
@@ -124,7 +69,7 @@ const	char *const __mdoc_macronames[MDOC_MAX] = {		 
 	/* LINTED */
 	"Dx",		"%Q",		"br",		"sp",
 	/* LINTED */
-	"%U"
+	"%U",		"Ta"
 	};
 
 const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {		 
@@ -151,16 +96,18 @@ static	struct mdoc_node *node_alloc(struct mdoc *, int
 				enum mdoct, enum mdoc_type);
 static	int		  node_append(struct mdoc *, 
 				struct mdoc_node *);
-static	int		  mdoc_ptext(struct mdoc *, int, char *);
-static	int		  mdoc_pmacro(struct mdoc *, int, char *);
-static	int		  macrowarn(struct mdoc *, int, const char *);
+static	int		  mdoc_ptext(struct mdoc *, int, char *, int);
+static	int		  mdoc_pmacro(struct mdoc *, int, char *, int);
+static	int		  mdoc_span_alloc(struct mdoc *, 
+				const struct tbl_span *);
 
 
 const struct mdoc_node *
 mdoc_node(const struct mdoc *m)
 {
 
-	return(MDOC_HALT & m->flags ? NULL : m->first);
+	assert( ! (MDOC_HALT & m->flags));
+	return(m->first);
 }
 
 
@@ -168,7 +115,8 @@ const struct mdoc_meta *
 mdoc_meta(const struct mdoc *m)
 {
 
-	return(MDOC_HALT & m->flags ? NULL : &m->meta);
+	assert( ! (MDOC_HALT & m->flags));
+	return(&m->meta);
 }
 
 
@@ -191,6 +139,8 @@ mdoc_free1(struct mdoc *mdoc)
 		free(mdoc->meta.arch);
 	if (mdoc->meta.vol)
 		free(mdoc->meta.vol);
+	if (mdoc->meta.msec)
+		free(mdoc->meta.msec);
 }
 
 
@@ -243,17 +193,15 @@ mdoc_free(struct mdoc *mdoc)
  * Allocate volatile and non-volatile parse resources.  
  */
 struct mdoc *
-mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
+mdoc_alloc(struct regset *regs, void *data, mandocmsg msg)
 {
 	struct mdoc	*p;
 
 	p = mandoc_calloc(1, sizeof(struct mdoc));
 
-	if (cb)
-		memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
-
+	p->msg = msg;
 	p->data = data;
-	p->pflags = pflags;
+	p->regs = regs;
 
 	mdoc_hash_init();
 	mdoc_alloc1(p);
@@ -269,113 +217,108 @@ int
 mdoc_endparse(struct mdoc *m)
 {
 
-	if (MDOC_HALT & m->flags)
-		return(0);
-	else if (mdoc_macroend(m))
+	assert( ! (MDOC_HALT & m->flags));
+	if (mdoc_macroend(m))
 		return(1);
 	m->flags |= MDOC_HALT;
 	return(0);
 }
 
-
-/*
- * Main parse routine.  Parses a single line -- really just hands off to
- * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
- */
 int
-mdoc_parseln(struct mdoc *m, int ln, char *buf)
+mdoc_addspan(struct mdoc *m, const struct tbl_span *sp)
 {
 
-	if (MDOC_HALT & m->flags)
-		return(0);
+	assert( ! (MDOC_HALT & m->flags));
 
-	m->flags |= MDOC_NEWLINE;
-	return('.' == *buf ? 
-			mdoc_pmacro(m, ln, buf) :
-			mdoc_ptext(m, ln, buf));
+	/* No text before an initial macro. */
+
+	if (SEC_NONE == m->lastnamed) {
+		mdoc_pmsg(m, sp->line, 0, MANDOCERR_NOTEXT);
+		return(1);
+	}
+
+	return(mdoc_span_alloc(m, sp));
 }
 
 
+/*
+ * Main parse routine.  Parses a single line -- really just hands off to
+ * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
+ */
 int
-mdoc_verr(struct mdoc *mdoc, int ln, int pos, 
-		const char *fmt, ...)
+mdoc_parseln(struct mdoc *m, int ln, char *buf, int offs)
 {
-	char		 buf[256];
-	va_list		 ap;
 
-	if (NULL == mdoc->cb.mdoc_err)
-		return(0);
+	assert( ! (MDOC_HALT & m->flags));
 
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
+	m->flags |= MDOC_NEWLINE;
 
-	return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
+	/*
+	 * Let the roff nS register switch SYNOPSIS mode early,
+	 * such that the parser knows at all times
+	 * whether this mode is on or off.
+	 * Note that this mode is also switched by the Sh macro.
+	 */
+	if (m->regs->regs[(int)REG_nS].set) {
+		if (m->regs->regs[(int)REG_nS].v.u)
+			m->flags |= MDOC_SYNOPSIS;
+		else
+			m->flags &= ~MDOC_SYNOPSIS;
+	}
+
+	return(('.' == buf[offs] || '\'' == buf[offs]) ? 
+			mdoc_pmacro(m, ln, buf, offs) :
+			mdoc_ptext(m, ln, buf, offs));
 }
 
 
 int
-mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
+mdoc_vmsg(struct mdoc *mdoc, enum mandocerr t, 
+		int ln, int pos, const char *fmt, ...)
 {
 	char		 buf[256];
 	va_list		 ap;
 
-	if (NULL == mdoc->cb.mdoc_warn)
-		return(0);
-
 	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
+	vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 	va_end(ap);
 
-	return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, buf));
+	return((*mdoc->msg)(t, mdoc->data, ln, pos, buf));
 }
 
 
 int
-mdoc_err(struct mdoc *m, int line, int pos, int iserr, enum merr type)
+mdoc_macro(MACRO_PROT_ARGS)
 {
-	const char	*p;
-
-	p = __mdoc_merrnames[(int)type];
-	assert(p);
-
-	if (iserr)
-		return(mdoc_verr(m, line, pos, p));
-
-	return(mdoc_vwarn(m, line, pos, p));
-}
-
-
-int
-mdoc_macro(struct mdoc *m, enum mdoct tok, 
-		int ln, int pp, int *pos, char *buf)
-{
 	assert(tok < MDOC_MAX);
 
 	/* If we're in the body, deny prologue calls. */
 
 	if (MDOC_PROLOGUE & mdoc_macros[tok].flags && 
-			MDOC_PBODY & m->flags)
-		return(mdoc_perr(m, ln, pp, EPROLBODY));
+			MDOC_PBODY & m->flags) {
+		mdoc_pmsg(m, line, ppos, MANDOCERR_BADBODY);
+		return(1);
+	}
 
 	/* If we're in the prologue, deny "body" macros.  */
 
 	if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && 
 			! (MDOC_PBODY & m->flags)) {
-		if ( ! mdoc_pwarn(m, ln, pp, EBODYPROL))
-			return(0);
+		mdoc_pmsg(m, line, ppos, MANDOCERR_BADPROLOG);
+		if (NULL == m->meta.msec)
+			m->meta.msec = mandoc_strdup("1");
 		if (NULL == m->meta.title)
-			m->meta.title = mandoc_strdup("unknown");
+			m->meta.title = mandoc_strdup("UNKNOWN");
 		if (NULL == m->meta.vol)
-			m->meta.vol = mandoc_strdup("local");
+			m->meta.vol = mandoc_strdup("LOCAL");
 		if (NULL == m->meta.os)
-			m->meta.os = mandoc_strdup("local");
+			m->meta.os = mandoc_strdup("LOCAL");
 		if (0 == m->meta.date)
 			m->meta.date = time(NULL);
 		m->flags |= MDOC_PBODY;
 	}
 
-	return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
+	return((*mdoc_macros[tok].fp)(m, tok, line, ppos, pos, buf));
 }
 
 
@@ -404,10 +347,25 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p)
 
 	p->parent->nchild++;
 
+	/*
+	 * Copy over the normalised-data pointer of our parent.  Not
+	 * everybody has one, but copying a null pointer is fine.
+	 */
+
+	switch (p->type) {
+	case (MDOC_BODY):
+		/* FALLTHROUGH */
+	case (MDOC_TAIL):
+		/* FALLTHROUGH */
+	case (MDOC_HEAD):
+		p->norm = p->parent->norm;
+		break;
+	default:
+		break;
+	}
+
 	if ( ! mdoc_valid_pre(mdoc, p))
 		return(0);
-	if ( ! mdoc_action_pre(mdoc, p))
-		return(0);
 
 	switch (p->type) {
 	case (MDOC_HEAD):
@@ -419,6 +377,8 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p)
 		p->parent->tail = p;
 		break;
 	case (MDOC_BODY):
+		if (p->end)
+			break;
 		assert(MDOC_BLOCK == p->parent->type);
 		p->parent->body = p;
 		break;
@@ -429,11 +389,11 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p)
 	mdoc->last = p;
 
 	switch (p->type) {
+	case (MDOC_TBL):
+		/* FALLTHROUGH */
 	case (MDOC_TEXT):
 		if ( ! mdoc_valid_post(mdoc))
 			return(0);
-		if ( ! mdoc_action_post(mdoc))
-			return(0);
 		break;
 	default:
 		break;
@@ -455,9 +415,17 @@ node_alloc(struct mdoc *m, int line, int pos, 
 	p->pos = pos;
 	p->tok = tok;
 	p->type = type;
+
+	/* Flag analysis. */
+
+	if (MDOC_SYNOPSIS & m->flags)
+		p->flags |= MDOC_SYNPRETTY;
+	else
+		p->flags &= ~MDOC_SYNPRETTY;
 	if (MDOC_NEWLINE & m->flags)
 		p->flags |= MDOC_LINE;
 	m->flags &= ~MDOC_NEWLINE;
+
 	return(p);
 }
 
@@ -505,6 +473,22 @@ mdoc_body_alloc(struct mdoc *m, int line, int pos, enu
 
 
 int
+mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
+		struct mdoc_node *body, enum mdoc_endbody end)
+{
+	struct mdoc_node *p;
+
+	p = node_alloc(m, line, pos, tok, MDOC_BODY);
+	p->pending = body;
+	p->end = end;
+	if ( ! node_append(m, p))
+		return(0);
+	m->next = MDOC_NEXT_SIBLING;
+	return(1);
+}
+
+
+int
 mdoc_block_alloc(struct mdoc *m, int line, int pos, 
 		enum mdoct tok, struct mdoc_arg *args)
 {
@@ -514,6 +498,21 @@ mdoc_block_alloc(struct mdoc *m, int line, int pos, 
 	p->args = args;
 	if (p->args)
 		(args->refcnt)++;
+
+	switch (tok) {
+	case (MDOC_Bd):
+		/* FALLTHROUGH */
+	case (MDOC_Bf):
+		/* FALLTHROUGH */
+	case (MDOC_Bl):
+		/* FALLTHROUGH */
+	case (MDOC_Rs):
+		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
+		break;
+	default:
+		break;
+	}
+
 	if ( ! node_append(m, p))
 		return(0);
 	m->next = MDOC_NEXT_CHILD;
@@ -531,13 +530,37 @@ mdoc_elem_alloc(struct mdoc *m, int line, int pos, 
 	p->args = args;
 	if (p->args)
 		(args->refcnt)++;
+
+	switch (tok) {
+	case (MDOC_An):
+		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
+		break;
+	default:
+		break;
+	}
+
 	if ( ! node_append(m, p))
 		return(0);
 	m->next = MDOC_NEXT_CHILD;
 	return(1);
 }
 
+static int
+mdoc_span_alloc(struct mdoc *m, const struct tbl_span *sp)
+{
+	struct mdoc_node *n;
 
+	n = node_alloc(m, sp->line, 0, MDOC_MAX, MDOC_TBL);
+	n->span = sp;
+
+	if ( ! node_append(m, n))
+		return(0);
+
+	m->next = MDOC_NEXT_SIBLING;
+	return(1);
+}
+
+
 int
 mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
 {
@@ -561,10 +584,12 @@ mdoc_word_alloc(struct mdoc *m, int line, int pos, con
 }
 
 
-void
+static void
 mdoc_node_free(struct mdoc_node *p)
 {
 
+	if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
+		free(p->norm);
 	if (p->string)
 		free(p->string);
 	if (p->args)
@@ -590,6 +615,8 @@ mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
 		n->parent->nchild--;
 		if (n->parent->child == n)
 			n->parent->child = n->prev ? n->prev : n->next;
+		if (n->parent->last == n)
+			n->parent->last = n->prev ? n->prev : NULL;
 	}
 
 	/* Adjust parse point, if applicable. */
@@ -629,122 +656,164 @@ mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
  * control character.
  */
 static int
-mdoc_ptext(struct mdoc *m, int line, char *buf)
+mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
 {
-	int		 i;
+	char		 *c, *ws, *end;
+	struct mdoc_node *n;
 
 	/* Ignore bogus comments. */
 
-	if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2])
-		return(mdoc_pwarn(m, line, 0, EBADCOMMENT));
+	if ('\\' == buf[offs] && 
+			'.' == buf[offs + 1] && 
+			'"' == buf[offs + 2]) {
+		mdoc_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
+		return(1);
+	}
 
 	/* No text before an initial macro. */
 
-	if (SEC_NONE == m->lastnamed)
-		return(mdoc_perr(m, line, 0, ETEXTPROL));
+	if (SEC_NONE == m->lastnamed) {
+		mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT);
+		return(1);
+	}
 
-	/* Literal just gets pulled in as-is. */
-	
-	if (MDOC_LITERAL & m->flags)
-		return(mdoc_word_alloc(m, line, 0, buf));
+	assert(m->last);
+	n = m->last;
 
-	/* Check for a blank line, which may also consist of spaces. */
+	/*
+	 * Divert directly to list processing if we're encountering a
+	 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
+	 * (a MDOC_BODY means it's already open, in which case we should
+	 * process within its context in the normal way).
+	 */
 
-	for (i = 0; ' ' == buf[i]; i++)
-		/* Skip to first non-space. */ ;
+	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
+			LIST_column == n->norm->Bl.type) {
+		/* `Bl' is open without any children. */
+		m->flags |= MDOC_FREECOL;
+		return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
+	}
 
-	if ('\0' == buf[i]) {
-		if ( ! mdoc_pwarn(m, line, 0, ENOBLANK))
-			return(0);
+	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
+			NULL != n->parent &&
+			MDOC_Bl == n->parent->tok &&
+			LIST_column == n->parent->norm->Bl.type) {
+		/* `Bl' has block-level `It' children. */
+		m->flags |= MDOC_FREECOL;
+		return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
+	}
 
+	/*
+	 * Search for the beginning of unescaped trailing whitespace (ws)
+	 * and for the first character not to be output (end).
+	 */
+
+	/* FIXME: replace with strcspn(). */
+	ws = NULL;
+	for (c = end = buf + offs; *c; c++) {
+		switch (*c) {
+		case '-':
+			if (mandoc_hyph(buf + offs, c))
+				*c = ASCII_HYPH;
+			ws = NULL;
+			break;
+		case ' ':
+			if (NULL == ws)
+				ws = c;
+			continue;
+		case '\t':
+			/*
+			 * Always warn about trailing tabs,
+			 * even outside literal context,
+			 * where they should be put on the next line.
+			 */
+			if (NULL == ws)
+				ws = c;
+			/*
+			 * Strip trailing tabs in literal context only;
+			 * outside, they affect the next line.
+			 */
+			if (MDOC_LITERAL & m->flags)
+				continue;
+			break;
+		case '\\':
+			/* Skip the escaped character, too, if any. */
+			if (c[1])
+				c++;
+			/* FALLTHROUGH */
+		default:
+			ws = NULL;
+			break;
+		}
+		end = c + 1;
+	}
+	*end = '\0';
+
+	if (ws)
+		mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
+
+	if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
+		mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
+
 		/*
-		 * Insert a `Pp' in the case of a blank line.  Technically,
+		 * Insert a `sp' in the case of a blank line.  Technically,
 		 * blank lines aren't allowed, but enough manuals assume this
 		 * behaviour that we want to work around it.
 		 */
-		if ( ! mdoc_elem_alloc(m, line, 0, MDOC_Pp, NULL))
+		if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
 			return(0);
 
 		m->next = MDOC_NEXT_SIBLING;
 		return(1);
 	}
 
-	/* 
-	 * Warn if the last un-escaped character is whitespace. Then
-	 * strip away the remaining spaces (tabs stay!).   
-	 */
-
-	i = (int)strlen(buf);
-	assert(i);
-
-	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
-		if (i > 1 && '\\' != buf[i - 2])
-			if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS))
-				return(0);
-
-		for (--i; i && ' ' == buf[i]; i--)
-			/* Spin back to non-space. */ ;
-
-		/* Jump ahead of escaped whitespace. */
-		i += '\\' == buf[i] ? 2 : 1;
-
-		buf[i] = '\0';
-	}
-
-	/* Allocate the whole word. */
-
-	if ( ! mdoc_word_alloc(m, line, 0, buf))
+	if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
 		return(0);
 
+	if (MDOC_LITERAL & m->flags)
+		return(1);
+
 	/*
 	 * End-of-sentence check.  If the last character is an unescaped
 	 * EOS character, then flag the node as being the end of a
 	 * sentence.  The front-end will know how to interpret this.
 	 */
 
-	assert(i);
+	assert(buf < end);
 
-	if (mandoc_eos(buf, (size_t)i))
+	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
 		m->last->flags |= MDOC_EOS;
 
 	return(1);
 }
 
 
-static int
-macrowarn(struct mdoc *m, int ln, const char *buf)
-{
-	if ( ! (MDOC_IGN_MACRO & m->pflags))
-		return(mdoc_verr(m, ln, 0, "unknown macro: %s%s", 
-				buf, strlen(buf) > 3 ? "..." : ""));
-	return(mdoc_vwarn(m, ln, 0, "unknown macro: %s%s",
-				buf, strlen(buf) > 3 ? "..." : ""));
-}
-
-
 /*
  * Parse a macro line, that is, a line beginning with the control
  * character.
  */
-int
-mdoc_pmacro(struct mdoc *m, int ln, char *buf)
+static int
+mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
 {
-	enum mdoct	tok;
-	int		i, j, sv;
-	char		mac[5];
+	enum mdoct	  tok;
+	int		  i, j, sv;
+	char		  mac[5];
+	struct mdoc_node *n;
 
 	/* Empty lines are ignored. */
 
-	if ('\0' == buf[1])
+	offs++;
+
+	if ('\0' == buf[offs])
 		return(1);
 
-	i = 1;
+	i = offs;
 
-	/* Accept whitespace after the initial control char. */
+	/* Accept tabs/whitespace after the initial control char. */
 
-	if (' ' == buf[i]) {
+	if (' ' == buf[i] || '\t' == buf[i]) {
 		i++;
-		while (buf[i] && ' ' == buf[i])
+		while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 			i++;
 		if ('\0' == buf[i])
 			return(1);
@@ -752,36 +821,28 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf)
 
 	sv = i;
 
-	/* Copy the first word into a nil-terminated buffer. */
+	/* 
+	 * Copy the first word into a nil-terminated buffer.
+	 * Stop copying when a tab, space, or eoln is encountered.
+	 */
 
-	for (j = 0; j < 4; j++, i++) {
-		if ('\0' == (mac[j] = buf[i]))
-			break;
-		else if (' ' == buf[i])
-			break;
+	j = 0;
+	while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
+		mac[j++] = buf[i++];
+	mac[j] = '\0';
 
-		/* Check for invalid characters. */
-
-		if (isgraph((u_char)buf[i]))
-			continue;
-		return(mdoc_perr(m, ln, i, EPRINT));
+	tok = (j > 1 || j < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
+	if (MDOC_MAX == tok) {
+		mdoc_vmsg(m, MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
+		return(1);
 	}
 
-	mac[j] = 0;
+	/* Disregard the first trailing tab, if applicable. */
 
-	if (j == 4 || j < 2) {
-		if ( ! macrowarn(m, ln, mac))
-			goto err;
-		return(1);
-	} 
-	
-	if (MDOC_MAX == (tok = mdoc_hash_find(mac))) {
-		if ( ! macrowarn(m, ln, mac))
-			goto err;
-		return(1);
-	}
+	if ('\t' == buf[i])
+		i++;
 
-	/* The macro is sane.  Jump to the next word. */
+	/* Jump to the next non-whitespace word. */
 
 	while (buf[i] && ' ' == buf[i])
 		i++;
@@ -792,13 +853,53 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf)
 	 */
 
 	if ('\0' == buf[i] && ' ' == buf[i - 1])
-		if ( ! mdoc_pwarn(m, ln, i - 1, ETAILWS))
+		mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
+
+	/*
+	 * If an initial macro or a list invocation, divert directly
+	 * into macro processing.
+	 */
+
+	if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
+		if ( ! mdoc_macro(m, tok, ln, sv, &i, buf)) 
 			goto err;
+		return(1);
+	}
 
-	/* 
-	 * Begin recursive parse sequence.  Since we're at the start of
-	 * the line, we don't need to do callable/parseable checks.
+	n = m->last;
+	assert(m->last);
+
+	/*
+	 * If the first macro of a `Bl -column', open an `It' block
+	 * context around the parsed macro.
 	 */
+
+	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
+			LIST_column == n->norm->Bl.type) {
+		m->flags |= MDOC_FREECOL;
+		if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
+			goto err;
+		return(1);
+	}
+
+	/*
+	 * If we're following a block-level `It' within a `Bl -column'
+	 * context (perhaps opened in the above block or in ptext()),
+	 * then open an `It' block context around the parsed macro.
+	 */
+
+	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
+			NULL != n->parent &&
+			MDOC_Bl == n->parent->tok &&
+			LIST_column == n->parent->norm->Bl.type) {
+		m->flags |= MDOC_FREECOL;
+		if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf)) 
+			goto err;
+		return(1);
+	}
+
+	/* Normal processing of a macro. */
+
 	if ( ! mdoc_macro(m, tok, ln, sv, &i, buf)) 
 		goto err;